Skip to content

4.3 DOS and PDOS

These are unexecuted teaching inputs and starting settings to test. Original figures are schematics, not computed results. Use licensed VASP and PAW data, replace every placeholder, record the executable version and validate convergence.

4.3.1 Model, units and provenance

Use eV for energy, Å for length and eV/Å for force; 1 kbar = 0.1 GPa. State normalization per atom, molecule, primitive cell or simulation cell. Record PAW identifiers, release, ZVAL, ENMAX and permitted hashes; never redistribute POTCAR. SCF convergence addresses the chosen electronic problem; convergence of the target property requires separate tests.

 DOS and PDOS — original conceptual schematic

Original schematic. Curves explain concepts; blank data areas await verified learner results. No calculation is claimed.

4.3.2 Worked case: procedure, interpretation and checks

Purpose and prerequisites

The density of states answers “how many states occur near this energy?” It deliberately discards the momentum coordinate. A nearly flat band can produce a large DOS because many k points occupy a narrow energy interval. A highly dispersive band may produce a modest DOS even when it crosses the Fermi level. Band and DOS plots are complementary views, not interchangeable calculations.

Use the same relaxed silicon structure as 4.2 Bands and effective masses. A magnetic transition metal is a useful second exercise, but it needs a independently converged spin-polarized ground state and appropriate occupations. Decide whether the plotted quantity will be per simulation cell, per formula unit, per atom, or per volume. These normalizations are not cosmetic: a doubled supercell doubles an extensive total DOS.

A practical static calculation

Make a fresh static property run on a dense regular mesh, not a line-mode path. For a ordinary three-dimensional bulk material with a suitable Γ-centered tetrahedral mesh, the following delta is a useful starting point:

Do not enter the EMIN/EMAX placeholders literally. First inspect the relevant eigenvalue interval and reference, then set a window covering the occupied manifold and requested unoccupied range. NEDOS controls the output energy sampling, not how many eigenstates VASP has actually computed. Increase NBANDS if the upper energy window lacks states. A dense printed grid cannot compensate for a coarse k mesh. NEDOS

The tetrahedron choice is not appropriate for every geometry or sparse mesh. For a Γ-only molecule or an initial test with insufficient tetrahedra, use an appropriate Gaussian broadening (ISMEAR=0) and examine its width. Do not confuse smoothing with physical temperature. Positive-order Methfessel–Paxton smearing is not a safe default for insulators. ISMEAR

A dense self-consistent run avoids ambiguity about whether the original density was itself k-converged. For ordinary LDA/GGA a separately converged density can instead be used for a fixed-density DOS run; this is a deliberate approximation to the same Hamiltonian, not an alternative route for hybrid DOS. Keep the magnetic state and functional unchanged.

Read the files, then interpret the chemistry

DOSCAR contains DOS and integrated DOS. Total DOS is extensive, in states/eV for the calculation cell; site projections carry their site normalization. Spin-polarized files have separate spin channels, and noncollinear projections have a different layout. Never parse all versions and spin modes with a single hard-coded column list. Use file headers or a maintained parser and check one block manually. A DOS collected during relaxation is not the final structure's clean static spectrum. DOSCAR

LORBIT=11 gives atom/orbital projections using the PAW projector machinery. These are useful labels of orbital character, but the atom-centered subspaces are not a unique partition of the entire electron density. The sum of plotted PDOS curves need not equal the total DOS exactly. The missing part is not automatically a numerical failure. Likewise, an integrated “d occupation” depends on the projection convention and is not a formal oxidation number. LORBIT

For a conventional collinear spin plot, it is common to draw the down-spin DOS below the axis. The negative ordinate is only a graphic convention; neither channel contains negative numbers of states. Explain whether the plotted energy zero is the SCF Fermi level or a band edge. For an insulator, the position of the Fermi level inside the gap is not an independently measured observable and can vary with the occupation convention.

Checks, pitfalls, and exercise

Compare the same spectrum for successively finer regular meshes at unchanged energy spacing and broadening. Track a small set of features: integrated occupied states, band-edge onset, one selected peak, and spin imbalance if present. A peak that moves or fragments with the mesh is not yet a reliable chemical fingerprint. If a narrow gap appears filled, reduce artificial broadening before declaring the system metallic. Cross-check the eigenvalue gap rather than reading the tail of a smoothed curve.

Exercise: plot total DOS, summed s projection, and summed p projection for silicon, then compare per-cell and per-atom normalization. Repeat with two k meshes and two output grids. Which changes alter the information, and which merely draw the same information more smoothly? Explain the PDOS sum discrepancy without inventing a “lost-electron” interpretation.

4.3.3 Unexecuted inputs and analysis scaffolds

These are unexecuted teaching inputs and starting settings to test. Original figures are schematics, not computed results. Use licensed VASP and PAW data, replace every placeholder, record the executable version and validate convergence.

4.3.3.1 Input block 1

# Static bulk DOS; retain a converged physical baseline
NSW    = 0
IBRION = -1
ISMEAR = -5
LORBIT = 11
NEDOS  = 3001
LCHARG = .TRUE.
LWAVE  = .FALSE.
# EMIN = <lower absolute energy in eV>
# EMAX = <upper absolute energy in eV>

4.3.5 Technical sources